CC BY
6
M. Yu. Knjazev
UiL OTS, Utrecht
CP-SELECTION IS NOT (ONLY) A LEXICAL PROPERTY: evidence from Russian
It was noted long ago that argumental CPs can appear in noncase positions alternating with PPs bearing presumably the same theta-role as the CP, as can be seen in (1).
(1a) Mary was not aware [CP that John was sick].
(1b) Mary was not aware [PP of the fact that John was sick].
(1c) *Mary was not aware [DP this fact].
According to an early intuition in [Rosenbaum 1967], such CPs as in (1a) were underlyingly embedded inside the PP, whose prepositional head was subsequently deleted. In what follows, I will use “deletion” as a descriptive term without committing myself to its theoretical implications.
As far as I am concerned, there has been no explicit characterization of the class of PP-selecting verb that allow preposition deletion, while it has been quite clear from the beginning that the process is lexically restricted. E.g., [Kuno 1973] mentions in a footnote that the adjective independent, unlike aware, disallows preposition deletion, as can be seen from (2).
(2a) That he is a good teacher is not independent from the fact that he is a good scholar.
(2b) *That he is a good teacher is not independent that he is a good scholar.
A null hypothesis about the contrast between (1a) and (2b) is that deletability of the PP is determined by the individual verb or, equivalently, be the unique theta-role borne by that-clause. According to this hypothesis, some theta-roles can be expressed by “bare” that-clauses along with PPs, while others can only be expressed by PPs. The next task for someone who adopts this hypothesis is to precisely characterize the two relevant classes of theta-roles.
CP-celection in not (only) a lexical property: evidence from Russian
I would like to suggest, however, that the null hypothesis in its strongest form, namely that deletability of the PP can be predicted solely on the basis of its unique theta-role, cannot be maintained. I provide evidence from Russian that deletability of the PP can depend on the argument/event structure of the predicate selecting this PP.
In Russian, CPs can be embedded inside PPs or oblique DPs selected by the predicate if these CPs are immediately preceded by a dummy demonstrative, as can be seen from the examples below. In some cases, just like in English, the preposition along with the demonstrative or the oblique-marked demonstrative can be deleted. At first glance, in many cases deletability of the demonstrative looks like a random lexical property, that is, it depends on the exact theta-role that a given verb assigns to the PP/DP. There are, however, interesting cases where the manipulation of the argument or event structure of a predicate influences the deletability of the demonstrative.
Consider, e.g., the case of obvinjat’ ‘accuse’. The reason argument of this verb is normally expressed by the PP headed by v ‘in’, as in (3a). Sometimes, as in (3b), the preposition along with the demonstrative is omitted. However, once the verb is passivized, the preposition has to stay there, as in (4). Podozrevat’ ‘suspect’ exhibits a similar pattern (example omitted). It is not clear how general this pattern is since a lot of transitive verbs taking propositional PP/DPs (e.g. rugat’ ‘scold’) seem to resist passivization but these examples do show that passivization can seriously degrade a verb’s ability to delete its PP/DP compement.
(3a) Oni ego obvinili v tom, cto on podpisal etot dokument.
they him accused in it.LOC that he signed this document
‘They accused him of assigning this document.’
(3b) ?Oni ego obvinili, cto on podpisal etot dokument.
(4a) On byl obvinen v tom, cto podpisal etot dokument.
he was accused in it.LOC that signed this document
‘They accused him of assigning this document.’
(4b) *On byl obvinen, cto podpisal etot dokument.
Another interesting case is provided by the verb rugat’ ‘scold’. The reason argument of this verb is normally expressed by the PP headed by za ‘for’, as in (5a). Sometimes, as in (5b), the preposition along with the demonstrative is omitted. However, once the verb is
M. Yu. Knjazev
perfectivized with a prefix, deletion of the PP becomes seriously degraded (6).
(5a) On rugal menja za to, cto ja opozdal.
he scolded me for it that I was.late
‘He scolded me for being late.’
(5b) On rugal menja, cto ja opozdal.
he scolded me that I was.late
‘He scolded me for being late.’
(6a) On obrugal menja za to, cto ja opozdal.
he scolded me for it that I was.late
‘He scolded me for being late.’
(6b) ??On obrugal menja, cto ja opozdal.
he scolded me that I was.late
‘He scolded me for being late.’
Similar facts seem to hold for (ot)blagodarit’ ‘thank’ and (ras)kritikovat’ ‘criticize’. The data do not appear crystal clear but
there are clear indications that perfectivization of the predicate can
affect deletability of its PP/DP argument in certain cases.
The nature and scope of the two phenomena discussed above
are still largely unclear and require further investigation but they seem to suggest that whether a given verb takes a that-clause as a complement may be at least partly determined by the argument/event structure of this verb and not only by the (unique) theta-role that it can assign to this clause, which is to say that CP-selection may not be entirely lexical, as has been mostly assumed.
Abbreviations
loc — locative case.
References
Kuno 1973 — S. Kuno. Constraints on internal clauses and sentential subjects // Linguistic Inquiry 4, 1973. P. 363-385.
Rosenbaum 1967 — P. S. Rosenbaum. The Grammar of English Predicate Complement Constructions. Cambridge, Mass.: MIT Press, 1967.
